USB 3.0/3.1/3.2 Gen 1 aka the kind that can do 5 gbps can only use cables up to 3m in length. Meanwhile Ethernet uses a similar twisted pair copper cable, but can do 10 gbps over 100m.
What gives, why is USB so limited in terms of cable length?
In: 413
“Because they were designed that way” isn’t very helpful, but is basically correct.
Ethernet cables are designed for connecting things throughout a building. In something like an office building, you need long run lengths (e.g connecting upper floors to the servers in the basement), and you need many km of cables so they can’t be too expensive. Long cable runs are subject to a lot of interference so pushing the data rate too high can lead to errors unless you physically bulk up the cable and voltages, or switch to fibre optics. Speed is theoretically nice, but realistically you don’t normally need to be transferring large files around anyway, and the bottleneck is usually the server or router having to deal with dozens or hundreds of users simultaneous requests and not the actual network speed. Reliability and practicality are more important, and ethernet is a good middle ground that’s cheap, easy to work with, fast enough for most purposes.
USB cables are for connecting peripherals to devices in the same room. Generally you don’t need more than a couple of meters, but you do want high speed for external storage, displays, cameras. You don’t have to worry too much about interference over a short run, nor do you have to have bulky cables and high voltages to counteract with resistance and signal drop off.
Can you do USB3 speeds at Ethernet distances? Absolutely! But either you need expensive, bulky cables and equipment or (much more sensibly) use fiber optics. Fiber is harder to work with, termination and splicing was a really fiddly manual process till relatively recently, and the switching equipment at both ends is expensive, but if you really need it that’s the route to take. Unfortunately it really isn’t suitable for “patch” connections between the wall and a device – a user can kick the crap out of an ethernet patch cable and it’ll still work, but bend a fiber the wrong way and it’s toast – so you often get fiber from the server room to the switch and then ethernet to the desk. Given that the ethernet to the switch is unlikely to be the bottleneck, this works pretty well outside niche applications involving large amounts of raw data, and TBH there you should probably be processing the data in place (using a proper local workstation or client/server model) rather than trying to drag data back and forth across a network.
Latest Answers